AF precision depends on aperture

Class A · 01-10-2009, 10:20 PM

Motivated by the question of whether or not the Tamron/Pentax 18-250 requires focus readjustment after zooming I did a test. My result is that the focus shifts slightly when the focal length is changed and hence zooming in, focusing, and then zooming out again won't give the optimal sharpness.

Anyhow, what I noticed is the following (with a K100D and a Tamron 18-250):

When I did the wide (18mm) shots at f/3.5, the AF system readjusted the focus obtained at 250mm.

When I repeated the same test at f/8, the AF system did not (or if at all, just very slightly) readjust the focus between focal lengths changes.

Since AF always happens wide open, this only leads to one conclusion: The AF system takes the target aperture into account. Obviously, at f/8 there is a larger DOF, so it kind of makes sense to avoid micro adjustments if the subject is deemed to be within the field considered to be sharp anyhow. At wide target apertures, the exact focus matters much more, therefore even slight adjustments are made.

For those who didn't know this already, it may be helpful to know that. One normally shoots focus charts at the widest aperture anyhow, but if for some reason one doesn't, one shouldn't expect the AF system to do the best it can.

The phenomenon may suggest that the most precise focusing could be obtained by setting the widest aperture possible for focussing and only after focus has been obtained set the intended shooting aperture. This only works, though, if the lens doesn't cause a focus shift when the aperture is changed.

blende8 · 01-11-2009, 02:17 AM

The more probable explanation is that the target at 18mm is quite different from that at 250mm.
Slight changes are probably chance.

Perhaps you are right, but this needs to be repeated several times in a controlled way with different targets and statistics to make sure that this is real.
It's not impossible. Lens information for Pentax lenses is stored in the camera.

Class A · 01-11-2009, 03:45 AM

Originally Posted by blende8

The more probable explanation is that the target at 18mm is quite different from that at 250mm.

For a non-constant focus zoom lens, which I believe the 18-250 is, this is definitely true. But this doesn't explain why no adjustment is made when the target aperture is f/8 or higher. The latter phenomenon can only be explained by a higher tolerance regarding focus for higher f-ratios.

Originally Posted by blende8

Perhaps you are right, but this needs to be repeated several times in a controlled way with different targets and statistics to make sure that this is real.

I fully agree. For instance, I checked it with my FA 50/1.4 and this lens never adjusted focus from a higher f-ratio to a lower f-ratio.

Maybe this phenomenon doesn't exist for primes (and potentially other lens types).

mithrandir · 01-11-2009, 08:37 AM

This brings to mind a quirk with my (and maybe all) K10Ds. I have noticed when my K10D starts to violently hunt back and forth for an AF, if I do a DOF preview then try to focus it will lock right on 90% of the time. I suspect now that after the preview the camera is remembering the DOF preview feedback and using it to set the AF. Any thoughts?

*jpzk* · 01-11-2009, 09:25 AM

I think this is a matter of which lens one uses.
For instance, I also have the 18-250 and it does behave that way under less-than-optimal lighting conditions; if the contrast is not high enough, the AF will start "hunting", same goes with my Sigma 100-300 F4,mounted on either my K10D and/or K20D.
On the DA*16-50 ... not such problem. AF very fast and absolutely no hunting ... it is a much luminous lens too. On the Sigma 10-20 ... same thing: no AF hunting... and so on.

I have had my fair share of missing pics while bird shooting under low light/mediocre contrast using the AF. More often than not, I will switch to MF during overcast days using long focals.
JP

*séamuis* · 01-11-2009, 09:29 AM

Originally Posted by mithrandir

This brings to mind a quirk with my (and maybe all) K10Ds. I have noticed when my K10D starts to violently hunt back and forth for an AF, if I do a DOF preview then try to focus it will lock right on 90% of the time. I suspect now that after the preview the camera is remembering the DOF preview feedback and using it to set the AF. Any thoughts?

that's quite an interesting theory. that should also be tested.

*joelovotti* · 01-11-2009, 02:23 PM

When you focus "wide open" at 250mm you are actually at f6.3. When you come back to 18mm you are now "wide open" at f3.5. The focus difference may be do to this aperture difference not the target aperture.

Class A · 01-11-2009, 02:36 PM

Originally Posted by mithrandir

I suspect now that after the preview the camera is remembering the DOF preview feedback and using it to set the AF. Any thoughts?

I believe "using the DOF preview feedback" for AF purposes would be way too sophisticated. I don't anything remotely like this is happening.

Is the described phenomenon really repeatable? Does switching the camera off and on again yield the same result (less hunting)?

If the phenomenon is indeed repeatable, I can only assume that the DOF preview accidentally resets some AF tolerance value (or similar). That seems unlikely though. I just did a test. After zooming out from a 250mm f/8 AF adjustment, a following DOF preview caused no AF readjustment. A change of the aperture from f/8 to f/3.5 did.

stevebrot · 01-11-2009, 03:44 PM

Back to the subject of your original post...

While it is usually safe to assume that a zoom lens is parfocal, this is not always the case. Focus shift while zooming is possible and probably fairly likely with an IF lens.

That the camera would refocus after zooming does not surprise me at all.

Steve

Class A · 01-11-2009, 03:46 PM

Originally Posted by joelovotti

When you focus "wide open" at 250mm you are actually at f6.3.
When you come back to 18mm you are now "wide open" at f3.5. The focus difference may be do to this aperture difference not the target aperture.

I fully agree with everything you say except for the last part of the last sentence.

If you change the last sentence to

The focus difference may be due to this aperture difference not the change in focal length.

then I agree. If the 18-250 had a drastic focus shift due to aperture change then that could mean that my tests don't show a focus difference induced by focal length change, but by aperture change.

I'm not experienced with focus shifts due to aperture changes, but I expect them to be small. When you focus at 18mm f/3.5 and then zoom in the image is rather blurred. I don't think this degree of misfocus is caused by a change in aperture. You might say that one should expect a focus at 18mm to not hold up at 250mm anymore, but my comparison shots show that, for a 18mm shot, the focus at 18mm is indeed better then the one obtained at 250mm.

Also note that when I focus at 250mm f/8, then zoom out to 18mm and then AF again, no readjustment is made, even though the AF now measures through f/3.5 again. Only by setting the target f-ratio from 8 to 3.5 I can cause the camera to readjust focus upon causing another AF. No aperture change occurred between the two AF initiations but in the first case no readjustment takes place but in the second it does.

Class A · 01-11-2009, 05:25 PM

Originally Posted by stevebrot

While it is usually safe to assume that a zoom lens is parfocal, this is not always the case.

Yes, it was more important in MF days for a zoom lens to be parfocal.
I believe a number of modern zoom lenses are varifocal.
As explained in What Is Inside a Zoom Lens? a parfocal lens requires a dedicated "compensator group". It is easy to imagine that superzooms such as the 18-250 are complex enough without such a group.

Originally Posted by stevebrot

That the camera would refocus after zooming does not surprise me at all.

Yes, but that it doesn't if the planned f-ratio is high enough did surprise me.

I did one more confirmation test. When manual focussing, the focus indicator (green hexagon) stays on over a much wider range when a high f-ratio is chosen, compared to setting a low f-ratio.

It makes perfect sense for the camera to work this way, but it also means that it won't try to put the centre of the focus on the subject when a high f-ratio is chosen. At least not with a lens like the 18-250.